Robotics and Computer Vision Laboratory Login  
  Robotics and Computer Vision Laboratory kaist logo
Archive Courses

Home  >  Research  >  Publications

[International Conference] Intelligent Grasping by Learning from Observation
Seong-Young Ko , Hanbyul Joo , Dong-Geol Choi , Hyeongwoo Kim , In So Kweon
The 5th International Conference on Ubiquitous Robots and Ambient Intelligence (URAI 2008) , November 2008
  2008URAI.pdf 2008URAI.pdf (1.7M) [79]
This paper explains the development of
intelligent grasping robot system to learn the manipulation
skill of the object and then to imitate its trajectory. The
system was developed so that it observes the posture and
the trace of the desired object, produces the trajectory for
5DoF robotic arm, and moves to mimic the trajectory
reversely. In order to extract the information of the object
in real time, the shape matching algorithm is applied to
stereo images. The center position of the object is
obtained at each image and these are utilized to calculate
the 3-D position in terms of the camera coordination
system. These points were converted in terms of the robot
coordination system, and the trajectory for robots motion
was generated by interpolating these points. The grasping
motion of the robotic arm is generated by using the start
position and the end position of the trajectory. It was
shown that the developed robotics system based on shape
matching method operates robustly.


Robotics and Computer Vision Laboratory
KAIST | Electrical Engineering | Contact Us | Sitemap